| prometheanfire | jamesdenton: you think ovn is still experimental in openstack? | 00:04 |
|---|---|---|
| *** DanyC has quit IRC | 00:07 | |
| *** gyee has quit IRC | 01:07 | |
| *** jamesdenton_wfh has joined #openstack-ansible | 01:20 | |
| *** jamesdenton_wfh has quit IRC | 01:34 | |
| *** rohit02 has joined #openstack-ansible | 02:18 | |
| *** rohit02 has quit IRC | 02:58 | |
| *** rh-jelabarre has quit IRC | 03:00 | |
| *** rohit02 has joined #openstack-ansible | 04:02 | |
| *** evrardjp has quit IRC | 04:36 | |
| *** evrardjp has joined #openstack-ansible | 04:37 | |
| *** ygk_12345 has joined #openstack-ansible | 05:20 | |
| *** udesale has joined #openstack-ansible | 05:22 | |
| *** udesale has quit IRC | 05:23 | |
| *** udesale has joined #openstack-ansible | 05:23 | |
| *** kleini has joined #openstack-ansible | 05:47 | |
| *** miloa has joined #openstack-ansible | 05:56 | |
| *** ygk_12345 has quit IRC | 06:05 | |
| *** rohit02 has quit IRC | 06:17 | |
| *** this10nly has joined #openstack-ansible | 06:17 | |
| *** ygk_12345 has joined #openstack-ansible | 06:46 | |
| CeeMac | morning | 06:55 |
| ygk_12345 | i am upgrading from stein to train. the os_placement playbook is throwing this error http://paste.openstack.org/show/791782/ | 06:56 |
| ygk_12345 | can someone help me please | 06:56 |
| ygk_12345 | should I include any placement specific configuration in the config or user_variable files ? | 06:57 |
| jrosser | ygk_12345: do you have hosts for the placemen service in your config file? https://github.com/openstack/openstack-ansible/blob/master/etc/openstack_deploy/openstack_user_config.yml.prod.example#L177-L184 | 07:16 |
| ygk_12345 | jrosser if I include them now in the config file, what playbooks should I rerun ? | 07:17 |
| *** rpittau|afk is now known as rpittau | 07:17 | |
| jrosser | do you have containers for the placement serivce on your infra nodes? | 07:17 |
| ygk_12345 | nope. can I include the present controllers ? | 07:17 |
| jrosser | i am hoping that the upgrade scripts take care of things here | 07:19 |
| jrosser | are you following the major upgrade guide for S->T ? | 07:19 |
| ygk_12345 | i am at the setup-openstack playbook after the previous upgrade scripts. now if I ionclude the placement hosts, should I rerun setup-hosts ? | 07:20 |
| ygk_12345 | yes I am following S-->T | 07:20 |
| ygk_12345 | https://docs.openstack.org/openstack-ansible/latest/admin/upgrades/major-upgrades.html#upgrading-manually | 07:20 |
| ygk_12345 | jrosser i have containers for placement service in place | 07:24 |
| *** thuydang has joined #openstack-ansible | 07:28 | |
| ygk_12345 | jrosser if I do not setup placement service, will it affect the entire train setup in anyway ? | 07:30 |
| jrosser | ygk_12345: oh yes you need it, it's mandatory | 07:35 |
| ygk_12345 | jrosser so whats the workaround in my case. I am running other playbooks meanwhile | 07:36 |
| jrosser | it used to be a part of nova, it tracks resource usage like CPU and RAM and so on, and is key to deciding how VM are scheduled to hosts | 07:36 |
| ygk_12345 | ok | 07:36 |
| jrosser | oh i see the trouble, you are following the upgrade guide for master branch | 07:37 |
| jrosser | you need this one https://docs.openstack.org/openstack-ansible/train/admin/upgrades/major-upgrades.html#upgrading-manually | 07:37 |
| ygk_12345 | jrosser so do I need 20.0.3 . I am using 20.0.2 now | 07:38 |
| jrosser | thats not really related, you want the latest tag of train to upgrade to | 07:39 |
| jrosser | but look closely at the URL for the upgrade guide | 07:39 |
| jrosser | one says 'latest', the other says 'train' | 07:39 |
| jrosser | so only the upgrade docs for 'train' give you the specifics of whats involved in a S->T upgrade | 07:40 |
| jrosser | which is why there is no mention of the placement migration in the 'latest' upgrade guide becasue that will eventually be the one relevant for T->U | 07:40 |
| ygk_12345 | i am using 20.0.2 at the moment from the 'git tag' result. | 07:41 |
| ygk_12345 | so is there a workaround in my case for the moment ? | 07:41 |
| jrosser | you need to follow the upgrade guide for S->T which i linked | 07:41 |
| ygk_12345 | ok | 07:41 |
| jrosser | there are specific things in there for placement | 07:41 |
| ygk_12345 | i am using that same guide now | 07:42 |
| ygk_12345 | openstack-ansible setup-openstack.yml -e 'placement_migrate_flag=true' | 07:42 |
| *** sshnaidm|afk is now known as sshnaidm | 07:42 | |
| ygk_12345 | ohhh I see now | 07:42 |
| jrosser | any also openstack-ansible "${SCRIPTS_PATH}/upgrade-utilities/deploy-config-changes.yml" -e 'placement_migrate_flag=true' | 07:43 |
| jrosser | *and | 07:43 |
| ygk_12345 | oh ok let me test that | 07:44 |
| *** tosky has joined #openstack-ansible | 08:00 | |
| *** aedc_ has joined #openstack-ansible | 08:02 | |
| *** aedc has quit IRC | 08:05 | |
| *** aedc_ has quit IRC | 08:11 | |
| *** DanyC has joined #openstack-ansible | 08:15 | |
| *** aedc has joined #openstack-ansible | 08:17 | |
| *** thuydang has quit IRC | 08:59 | |
| *** ygk_12345 has quit IRC | 09:22 | |
| *** ygk_12345 has joined #openstack-ansible | 09:38 | |
| ygk_12345 | hi all I have upgraded from rocky----------stein------train successfully. but the horizon is not allowing me to select any domain from the GUI login page. is this a bug ? | 09:41 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-tests master: Add ansible openstack collection https://review.opendev.org/718362 | 09:45 |
| jrosser | noonedeadpunk: ansible collections WIP ^ | 09:46 |
| *** jbadiapa has joined #openstack-ansible | 09:59 | |
| *** rpittau is now known as rpittau|bbl | 10:24 | |
| *** cshen has joined #openstack-ansible | 10:45 | |
| noonedeadpunk | this is neat | 10:56 |
| *** sshnaidm is now known as sshnaidm|afk | 10:56 | |
| jrosser | noonedeadpunk: i think that will work as a first attempt | 11:02 |
| jrosser | maybe i make a DNM change to some repo with functional tests to see what happens..... | 11:02 |
| jrosser | but don't know how yet we could make a depends-on change for patches to the ansible collections repo itself | 11:02 |
| jrosser | and also the rabbitmq collection is not yet published | 11:03 |
| jrosser | and it seems impossible to install a collection from git | 11:03 |
| *** this10nly has quit IRC | 11:04 | |
| *** jamesdenton_wfh has joined #openstack-ansible | 11:05 | |
| noonedeadpunk | Huh, I think it is? https://github.com/ansible-collections/rabbitmq | 11:16 |
| noonedeadpunk | or you mean it's not in galaxy yet? | 11:16 |
| jrosser | yes i don't think it's been released to galaxy | 11:23 |
| jrosser | which then took me down a rabbit hole of how to install from git | 11:23 |
| noonedeadpunk | jamesdenton_wfh: if you have some environemtns to test swift at, maybe you can take a look why auditors are still not happy with? | 11:24 |
| noonedeadpunk | I can spawn a sandbox otherwise | 11:24 |
| *** this10nly has joined #openstack-ansible | 11:25 | |
| jamesdenton_wfh | noonedeadpunk did you see something in the log? | 11:25 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/bfd3249a28dc481294afbbfe89336376/log/logs/host/swift-account-auditor.service.journal-21-41-40.log.txt#7833 | 11:25 |
| jamesdenton_wfh | oh yes, that's actually the '1%' issue | 11:25 |
| noonedeadpunk | oh, is it still the case | 11:25 |
| jamesdenton_wfh | appears so, yes | 11:25 |
| noonedeadpunk | I can recall we were fixing and then reverting... | 11:26 |
| jamesdenton_wfh | indeed | 11:26 |
| jamesdenton_wfh | so maybe revert the revert? :D | 11:26 |
| noonedeadpunk | there was a bug report where they were saying they are going to fix it I guess? | 11:26 |
| jamesdenton_wfh | sounds right. think i've closed all those tabs. will need to find it again | 11:27 |
| noonedeadpunk | I guess that's the revert patch https://review.opendev.org/#/c/683102/1/defaults/main.yml | 11:29 |
| noonedeadpunk | oh, yeah, it was wrong revert.... | 11:30 |
| noonedeadpunk | But I think swift has fixed that here? https://review.opendev.org/#/c/685455/ | 11:31 |
| noonedeadpunk | I guess they've dropepd that along with six drop | 11:32 |
| noonedeadpunk | huh, no it's still there | 11:33 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Revert "Revert "Update fallocate_reserve to work with py3"" https://review.opendev.org/718400 | 11:38 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Revert "Revert "Update fallocate_reserve to work with py3"" https://review.opendev.org/718400 | 11:40 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Revert "Revert "Update fallocate_reserve to work with py3"" https://review.opendev.org/718400 | 11:40 |
| *** thuydang has joined #openstack-ansible | 11:42 | |
| *** rh-jelabarre has joined #openstack-ansible | 11:45 | |
| *** cshen has quit IRC | 11:47 | |
| *** cshen has joined #openstack-ansible | 11:49 | |
| *** DanyC has quit IRC | 11:55 | |
| *** DanyC has joined #openstack-ansible | 12:02 | |
| *** sshnaidm|afk is now known as sshnaidm | 12:19 | |
| *** ygk_12345 has quit IRC | 12:19 | |
| *** rpittau|bbl is now known as rpittau | 12:23 | |
| *** spotz has joined #openstack-ansible | 12:44 | |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Add ansible openstack collection https://review.opendev.org/718431 | 12:55 |
| *** spatel has joined #openstack-ansible | 12:58 | |
| *** spatel has quit IRC | 12:59 | |
| *** gshippey has joined #openstack-ansible | 13:02 | |
| *** udesale_ has joined #openstack-ansible | 13:18 | |
| *** udesale has quit IRC | 13:20 | |
| jamesdenton | prometheanfire The implementation of OVN is "experimental" in OSA, and it hasn't been touched in a while. But OVN itself is propbably fine | 13:30 |
| jamesdenton | i need to revisit it | 13:30 |
| *** ygk_12345 has joined #openstack-ansible | 13:35 | |
| noonedeadpunk | I think we've left neutron for you :p | 13:46 |
| jamesdenton_wfh | :D | 13:47 |
| jamesdenton_wfh | and i have failed you. my bad | 13:47 |
| *** ygk_12345 has left #openstack-ansible | 13:47 | |
| noonedeadpunk | I don't think you did | 13:47 |
| openstackgerrit | Amy Marrich (spotz) proposed openstack/openstack-ansible master: Grammar Cleanup - Maintenance Tasks https://review.opendev.org/718458 | 13:56 |
| *** this10nly has quit IRC | 14:13 | |
| bjoernt | Can i get a +2 on https://review.opendev.org/#/c/718278/ | 14:20 |
| mgariepy | bjoernt, lgtm | 14:21 |
| bjoernt | Do we know when 20.0.3 arrives as ironic is broken. currently ? | 14:22 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-os_glance master: [DNM] test ansible collections https://review.opendev.org/718473 | 14:29 |
| jrosser | bjoernt: out of interest which platform do you need virtualenv 20 for on train? | 14:30 |
| bjoernt | I already ran into this in stein where virtualenv was upgraded to 20 | 14:31 |
| jrosser | thats not quite what i mean - which OS is providing virtualenv 20 now? | 14:34 |
| jrosser | bjoernt: also i think 20.1.0 is now the latest tag on stable/train | 14:35 |
| bjoernt | oh thanks | 14:36 |
| bjoernt | i checked early last week and it wasnt there, lol | 14:36 |
| bjoernt | bionic stock is still 15 but it got upgraded to 20 throughout the install | 14:39 |
| kleini | I am testing the upgrade to Train. https://docs.openstack.org/openstack-ansible/train/admin/upgrades/major-upgrades.html <- this guide mentions a version 20.0.3, which does not seem to exist in the Git repository. Instead a tag 20.1.0 is available. I used now stable/train. Furthermore the upgrade guide mentions a "cleanup of the pip.conf file" with scripts/pip-conf-removal.yml. That playbook does not exist in stable/train. | 14:42 |
| kleini | What is the correct way to proceed here? | 14:42 |
| *** gouthamr has left #openstack-ansible | 14:42 | |
| *** gouthamr has joined #openstack-ansible | 14:42 | |
| jrosser | bjoernt: thats still surprising...... https://github.com/openstack/requirements/blob/stable/train/upper-constraints.txt#L678 | 14:43 |
| *** cshen has quit IRC | 14:43 | |
| bjoernt | yeah I need to check at which phase that happend | 14:43 |
| *** udesale_ has quit IRC | 14:46 | |
| jrosser | kleini: i think the pip.conf cleanup is only relevant to the R->S upgrade | 14:47 |
| jrosser | that is a bug in the documentation | 14:47 |
| kleini | Found the pip.conf cleanup playbook on stable/stein. Running that said "ok" for every existing container. So documentation issue. Thanks. Should I try to submit a fix for that? | 14:48 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible stable/train: Drop pip-conf-removal.yml from docs https://review.opendev.org/718480 | 14:48 |
| jrosser | ^ that should do it - that doc section was already fixed on master but not backported to train | 14:49 |
| jrosser | the issue with the version number on the docs is likely that that docs generation inserts what the next release *should* be, i.e 20.0.2 -> 20.0.3 | 14:50 |
| jrosser | i would expect that gets fixed next time a patch lands on stable/train and the documentation is rebuilt | 14:51 |
| *** cshen has joined #openstack-ansible | 15:03 | |
| kleini | okay, thank you once again | 15:05 |
| *** gyee has joined #openstack-ansible | 15:13 | |
| *** thuydang has quit IRC | 15:20 | |
| prometheanfire | jamesdenton_wfh: well, given our focus on low latency high pps stuff, we might just go with pure sr-iov (a ton of broadcast udp) | 15:22 |
| openstackgerrit | Amy Marrich (spotz) proposed openstack/openstack-ansible master: Grammar Cleanup - Maintenance Tasks https://review.opendev.org/718491 | 15:30 |
| CeeMac | jamesdenton: are you about? | 15:34 |
| openstackgerrit | Amy Marrich (spotz) proposed openstack/openstack-ansible master: Grammar Cleanup - Maintenance Tasks https://review.opendev.org/718458 | 15:35 |
| noonedeadpunk | spotz: oh, btw, do we need to re-work our contributing part? | 15:36 |
| jrosser | @spotz if you are having a spring clean of docs i made a bit of a mess of the formatting here | 15:36 |
| jrosser | https://docs.openstack.org/openstack-ansible/latest/user/limited-connectivity/index.html | 15:36 |
| jrosser | "Practice B" has all got mangled there | 15:36 |
| spotz | jrosser Ok I'll get that next. I just realized I only fix other people's patches or do governance patches so was feeling guilty | 15:37 |
| jrosser | awesome - thanks | 15:37 |
| jamesdenton_wfh | prometheanfire sounds reasonable | 15:39 |
| spotz | And if there are docs grammar stuff just tag me, in the vein of not nitpicking on reviews some stuff has gotten through. If I do the review I tend to fix in a commit on it but I'm not always a reviewer on them | 15:39 |
| kleini | Are there any guides, how to upgrade a Ceph cluster deployed along with OSA, when upgrading S->T? | 15:42 |
| CeeMac | jamesdenton_wfh: quick question, can you have multiple routers connected to a network thats shared with rbac | 15:43 |
| jamesdenton_wfh | a shared external network? as far as I know, you can | 15:43 |
| CeeMac | access_as_shared | 15:43 |
| jamesdenton_wfh | so in that case, while technically yes, i would expect a single router (as the gateway for the subnet) | 15:46 |
| *** miloa has quit IRC | 15:46 | |
| CeeMac | well, so far my experience has been technically no, lol | 15:46 |
| jamesdenton_wfh | lol | 15:46 |
| CeeMac | basically what I've got is a master project with a firewall appliance (you'll remember my foray into OPNsense a while back) | 15:47 |
| CeeMac | then I have "sub projects" which all link back to servers and the FW in teh master project for access to the internet, core services etc | 15:47 |
| jamesdenton_wfh | What i mean by that, is I *think* you can attach multiple routers by assigning specific ports rather than relying on neutron to assign a port as gateway. | 15:47 |
| CeeMac | the Master project has the OPNsense firewall attached to the shared network | 15:47 |
| CeeMac | the first sub project attached a router to the network with default gateway pointing to the OPNSense | 15:48 |
| CeeMac | when I've tried to do the same at the next 2 sub-projects - no subnets available, even though there is, I can see it and everything :'( | 15:48 |
| CeeMac | ah, using the old cli? | 15:48 |
| CeeMac | i'd thought about that as my next step | 15:48 |
| jamesdenton_wfh | so, for this shared network/subnet, how did you attach the router? | 15:49 |
| CeeMac | horizon | 15:49 |
| CeeMac | add interface, select from dropdown, boom. | 15:49 |
| CeeMac | no boom, in the other projects | 15:49 |
| jamesdenton_wfh | ok. so in that case, i would think that the router took on the IP you have set for the gateway of that subnet | 15:50 |
| jamesdenton_wfh | can you confirm,? | 15:50 |
| CeeMac | let me check one sec | 15:50 |
| CeeMac | i created the network and the subnet in the master project | 15:50 |
| CeeMac | ok, the subnet was created with no gateway IP | 15:51 |
| jamesdenton_wfh | hmm | 15:51 |
| jamesdenton_wfh | so what IP did the router get? | 15:51 |
| CeeMac | the router in the first sub project has no external gateway | 15:51 |
| jrosser | this feels a little circular as the subnet gateway ip would be the ip of the master router you want to create later | 15:51 |
| CeeMac | there is no master router | 15:52 |
| jrosser | aaahhh | 15:52 |
| CeeMac | the OPNsense is doing that job | 15:52 |
| jamesdenton_wfh | so what is the purpose of the router, exactly? the neutron router? | 15:52 |
| CeeMac | for each of the subprojects I need a router to tell the traffic from the local network where to go to reach the shared core services and internet | 15:52 |
| CeeMac | so, router, was my thought | 15:53 |
| CeeMac | each sub project was going to have its own router attached to the shared 'transport' network and its dedicated project network that the instances would sit on | 15:53 |
| jamesdenton_wfh | ok, so each project will have it's own respective network, and the router will be the gateway for that network? and the router itself needs a leg in the network with the opnsense firewall? | 15:53 |
| jamesdenton_wfh | yeah ok | 15:53 |
| CeeMac | the project network default gateway would be on the router, and the router has a static route to the opnsense | 15:53 |
| CeeMac | yes, just so | 15:54 |
| jamesdenton_wfh | would the router ALSO be connected to some external network for internet access? | 15:54 |
| CeeMac | no | 15:54 |
| CeeMac | only internet access is via opnsense | 15:54 |
| jamesdenton_wfh | got it | 15:54 |
| CeeMac | are you thinking access_as_external and disable NAT? | 15:54 |
| jamesdenton_wfh | you could do that, but you would need to implement some static routes on that opnsense | 15:54 |
| CeeMac | and use that network as the routers gateway network? | 15:55 |
| jamesdenton_wfh | yes | 15:55 |
| jamesdenton_wfh | you could do static routes or bgp if you're feeling froggy | 15:55 |
| CeeMac | i have static routes on the opnsense for the subproject networks anyhow | 15:55 |
| CeeMac | bgp could be fun at some point | 15:55 |
| jamesdenton_wfh | what's the nexthop address for those? it would need to be the respective project networks | 15:55 |
| CeeMac | can you have the same network with different RBAC> | 15:55 |
| CeeMac | like I can keep the one project access_as_shared, and the next ones I'm working on as access_as_external? | 15:56 |
| CeeMac | dont want to break the current one | 15:56 |
| CeeMac | well, the static route on the OPNSense right now was pointing at the router for the one sub project that is working | 15:56 |
| jamesdenton_wfh | if you want to keep things AS-IS, you should create a port for each router on the transport network, and use 'openstack router add port <router> <transport_port>' for each router. then configure static routes in the routers using 'openstack router set <router> --route destination=<subnet>,gateway=<ip-address>' | 15:57 |
| jamesdenton_wfh | i'm really curious to know what neutron assigned to the router interface if the gateway wasn't set in the subnet | 15:57 |
| CeeMac | ok, i might give that a go for now. i need to break my relience on horizon anyway :) | 15:57 |
| jamesdenton_wfh | take off the training wheels :D | 15:58 |
| CeeMac | the router that is there already? | 15:58 |
| jrosser | noonedeadpunk: the collections job on glance is passing, bar the rabbitmq failure we expect in the upgrade jobs | 15:58 |
| CeeMac | but they're so convenient :D | 15:58 |
| CeeMac | i told it what IP to give | 15:58 |
| jrosser | i can see in the ARA report that it's picked up the os_<> modules from the collection rather than the one inside ansible | 15:58 |
| jamesdenton_wfh | CeeMac Ok then. Yeah, try thru the CLI and see if it behaves differenttly | 15:59 |
| CeeMac | sure, will do, thanks for the advise, as always :D | 15:59 |
| jamesdenton_wfh | that is, trying to add other routers to the same network | 15:59 |
| jamesdenton_wfh | happy to help | 15:59 |
| CeeMac | yep, i'll get on that now and report back | 16:00 |
| *** sshnaidm is now known as sshnaidm|afk | 16:04 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_horizon master: [doc] Add description to horizon_images_upload_mode https://review.opendev.org/718509 | 16:09 |
| *** rpittau is now known as rpittau|afk | 16:23 | |
| CeeMac | jamesdenton_wfh: so. that didn't go so well. didnt get any errors back from adding the port to the router, showing router lists no interfaces, and adding route fails 'the nexthop is not connected with router' | 16:25 |
| CeeMac | so I need to add the subnet to the router also> | 16:25 |
| jamesdenton_wfh | how did you create the port? | 16:30 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_glance master: Update glance_store variables https://review.opendev.org/718513 | 16:31 |
| CeeMac | openstack port create --network <network> --fixed-ip subnet=<transport-subnet>,ip-address=<transport-ip> --project <sub-project> <portname> | 16:35 |
| CeeMac | so, just for the giggles I was able to attach the port to the router through horizon as admin using 'device id', and the route now takes, however the port is showing DOWN. I've tried setting it --enable, still DOWN | 16:36 |
| *** DanyC has quit IRC | 16:36 | |
| CeeMac | created the port as 'member' of the sub-project incidentally | 16:36 |
| *** evrardjp has quit IRC | 16:37 | |
| jamesdenton_wfh | that should be OK | 16:37 |
| *** evrardjp has joined #openstack-ansible | 16:37 | |
| CeeMac | the cli sommands seem odd | 16:37 |
| *** cshen has quit IRC | 16:37 | |
| CeeMac | when you show the router once the interface is added, it is listed as subnet-id, ip-address, port-id | 16:38 |
| CeeMac | yet the commands to add port and subnet are seperate in the cli | 16:38 |
| jamesdenton_wfh | yes, so it's one or the other | 16:38 |
| CeeMac | is the subnet inferred from the 'port add' | 16:38 |
| jamesdenton_wfh | if you do add subnet, then neutron will create a port and give it a fixed ip of the subnet's gateway | 16:39 |
| jamesdenton_wfh | if you do add port, then you control the ip it gets | 16:39 |
| CeeMac | oh.... | 16:39 |
| CeeMac | that didn't happen from my 'member' user, command completed and returned nothing | 16:39 |
| CeeMac | then router show, had no interface listed | 16:39 |
| CeeMac | same as for 'port add' | 16:39 |
| jamesdenton_wfh | does it look different as an admin user? | 16:40 |
| CeeMac | so maybe because i created the network as admin context its limiting what access users have | 16:40 |
| CeeMac | ports still shows down as admin user | 16:42 |
| jamesdenton_wfh | does the router show output look any different? | 16:42 |
| jamesdenton_wfh | can you pop into the qrouter namespace and see if the interface is there? | 16:42 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: [doc] Add line to distribute nova keys https://review.opendev.org/718515 | 16:43 |
| jamesdenton_wfh | also, is this linuxbridge? ovs? ovs+dvr? | 16:43 |
| CeeMac | ovs | 16:43 |
| jamesdenton_wfh | with L3 HA or DVR? | 16:43 |
| CeeMac | and i do see the interface listed now in the router since i attached it as admin, its just showing as down, so I don't think its plugged correctly | 16:43 |
| CeeMac | legacy HA | 16:43 |
| CeeMac | prepared for DVR, but not using just now | 16:43 |
| CeeMac | used the HA-tool thingy | 16:44 |
| CeeMac | in the deployment | 16:44 |
| CeeMac | so, showing router as admin lists the same port I attached, and the 2 ha-network ports | 16:45 |
| jamesdenton_wfh | i'm not sure DVR will properly support this scenario. but, maybe. | 16:45 |
| CeeMac | yeah, i'm not using dvr just now anyhow | 16:45 |
| jamesdenton_wfh | so, disregard port state. check the namespace | 16:45 |
| CeeMac | i think i'll maybe adapt/adopt vyos or something in the future and keep it all VM based routing | 16:45 |
| CeeMac | one sec | 16:45 |
| jamesdenton_wfh | or give your opnsense device an interface in each project network and call it a day | 16:46 |
| CeeMac | would need to share the project network back to the core project | 16:47 |
| CeeMac | ? | 16:47 |
| CeeMac | that could work I guess | 16:47 |
| CeeMac | wasn't sure if there was a hard limit to the number of interfaces an instance can have | 16:47 |
| CeeMac | ok, so, no, the IP / interface does not show up in the qrouter ns | 16:49 |
| CeeMac | just the ha port | 16:49 |
| jamesdenton_wfh | well, there is a KVM limit, and there's a VM limit | 16:50 |
| CeeMac | yeah | 16:50 |
| jamesdenton_wfh | yeah, not really sure what's wrong there. you might find something in the l3 agent log | 16:50 |
| CeeMac | i'm used to working with a hard limit of 10 in VMware so don't really view multiple NICs as an optio | 16:50 |
| jamesdenton_wfh | how many would you need? | 16:51 |
| CeeMac | depends i guess | 16:51 |
| *** DanyC has joined #openstack-ansible | 16:52 | |
| CeeMac | so there are 3 right now (WAN, LAN, Tranport) | 16:52 |
| CeeMac | then 1 for each sub-project of which there are currently 5 | 16:52 |
| CeeMac | but we have more business units taking interest in the project so that could extend out | 16:52 |
| jamesdenton_wfh | noonedeadpunk I didn't realize 'content' wasn't a new key you created out of thin air: https://bugs.launchpad.net/openstack-ansible/+bug/1829372 | 16:53 |
| openstack | Launchpad bug 1829372 in openstack-ansible "Usage of overrides as the only source for config_template " [Undecided,Invalid] | 16:53 |
| noonedeadpunk | not sure I got what you mean | 16:57 |
| noonedeadpunk | we kinda already use it I guess, ie https://opendev.org/openstack/openstack-ansible-os_glance/src/branch/master/tasks/glance_post_install.yml#L87 | 16:57 |
| noonedeadpunk | or https://opendev.org/openstack/openstack-ansible-os_zun/src/branch/master/tasks/zun_compute.yml#L82-L95 | 16:58 |
| noonedeadpunk | I think it was there for ages? | 17:02 |
| noonedeadpunk | oh, wait, this bug was a bit about smth else... | 17:02 |
| noonedeadpunk | yeah youre right | 17:03 |
| noonedeadpunk | jamesdenton_wfh: btw, does https://bugs.launchpad.net/openstack-ansible/+bug/1859159 look fair to you? | 17:05 |
| openstack | Launchpad bug 1859159 in openstack-ansible "os_swift train incorect default systemctl configuration" [Undecided,New] | 17:05 |
| CeeMac | jamesdenton_wfh: for the record, recreating the router and creating the port with device-id of the router as admin didnt make a difference to the outcome, still no port in qrouter ns | 17:07 |
| CeeMac | can't see any obvious record in l3-agent logs either | 17:08 |
| *** cshen has joined #openstack-ansible | 17:09 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Fix replicator services https://review.opendev.org/718524 | 17:13 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Fix replicator services https://review.opendev.org/718524 | 17:13 |
| *** cshen has quit IRC | 17:13 | |
| jamesdenton_wfh | noonedeadpunk I will have to look at my env to see how it's configured. | 17:25 |
| jamesdenton_wfh | CeeMac When you create the port, don't assign the device id. Just create it and let neutron handle device assignment when you attach | 17:26 |
| noonedeadpunk | jamesdenton_wfh: according to CI logs we're really doing it wrong... | 17:26 |
| noonedeadpunk | but yeah, if you have somewhere to check this would be great | 17:27 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-tests master: Synchronise move to openstack ansible collection to all roles https://review.opendev.org/718527 | 17:27 |
| openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible-tests master: Synchronise move to openstack ansible collection to all roles https://review.opendev.org/718527 | 17:27 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: [doc] Add Debian buster to the list of supported OS https://review.opendev.org/718528 | 17:31 |
| noonedeadpunk | jrosser: oh, wow, you decided to use openstack from collections as well - cool | 17:40 |
| jrosser | noonedeadpunk: yeah, well the rabbit one isn’t ready yet | 17:41 |
| noonedeadpunk | and we can't get it just from git? | 17:41 |
| jrosser | and I figure if we can get using the collection for openstack then we decouple from all the broken versions | 17:41 |
| jrosser | I’ve not tried that yet, just dropping the git repo on the collections path | 17:42 |
| jrosser | but that will be necessary in zuul for depends-on | 17:42 |
| noonedeadpunk | ok, I see | 17:43 |
| jrosser | noonedeadpunk: the todo list for rabbitmq collections is here pretty much https://github.com/ansible-collections/rabbitmq/issues | 17:46 |
| noonedeadpunk | that's.... pretty massive I'd say | 17:47 |
| noonedeadpunk | We probably should work on it one day | 17:48 |
| jrosser | Jesse was going to spend some time Friday and Monday looking at it, if we want to help out that would be a good time | 17:49 |
| *** cshen has joined #openstack-ansible | 17:59 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_glance master: Make possible to run glance-api not with uwsgi https://review.opendev.org/718530 | 18:00 |
| jrosser | mnaser: there’s other ugly with glance interoperable import https://bugzilla.redhat.com/show_bug.cgi?id=1552583 | 18:03 |
| openstack | bugzilla.redhat.com bug 1552583 in openstack-tripleo-heat-templates "[Glance] Interoperable image import should be disabled by default on 3 node HA setup" [Urgent,Closed: errata] - Assigned to pdeore | 18:03 |
| *** cshen has quit IRC | 18:04 | |
| noonedeadpunk | jrosser: yeah, was kinda aware of thet, so I'm now fixing haproxy balancing mode | 18:06 |
| jrosser | does that help? api endpoint sure but does taskflow use rpc? | 18:07 |
| noonedeadpunk | not sure about rpc.... | 18:11 |
| CeeMac | jamesdenton_wfh: is that not what I did the first time though? | 18:13 |
| CeeMac | i'll try playing around again in the morning | 18:13 |
| jamesdenton_wfh | i'm not sure, sorry | 18:15 |
| noonedeadpunk | jrosser: ok, it's not but will be.... | 18:15 |
| jrosser | I had quite a long look at interoperable import | 18:16 |
| jrosser | and eventually didn’t do it | 18:16 |
| noonedeadpunk | so like out of their irc "[21:14] <abhishekk> noonedeadpunk, it is on the table, we might change it to use RPC in near future" | 18:19 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Change glance haproxy balancing method https://review.opendev.org/718533 | 18:21 |
| noonedeadpunk | jrosser: can you check this one? https://review.opendev.org/#/c/607474/ | 18:33 |
| *** DanyC has quit IRC | 18:35 | |
| jrosser | gshippey: can you take a look at that? | 18:35 |
| jrosser | noonedeadpunk: gshippey is doing lots of stuff on federation right now so we’ll check that patch over tomorrow | 18:36 |
| noonedeadpunk | ok, nice, thanks | 18:36 |
| *** thuydang has joined #openstack-ansible | 18:42 | |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_glance master: Make possible to run glance-api not with uwsgi https://review.opendev.org/718530 | 18:43 |
| openstackgerrit | Merged openstack/openstack-ansible stable/train: Drop pip-conf-removal.yml from docs https://review.opendev.org/718480 | 18:44 |
| openstackgerrit | Amy Marrich (spotz) proposed openstack/openstack-ansible master: Grammar Cleanup - Maintenance Tasks https://review.opendev.org/718458 | 19:00 |
| spotz | jrosser: there was an extra : in the line above the code | 19:01 |
| jrosser | i was suspicious of that - but wondered if it was some markup magic | 19:01 |
| openstackgerrit | Merged openstack/openstack-ansible-repo_build master: Retire openstack-ansible-repo_build role https://review.opendev.org/716389 | 19:03 |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install master: Retire pip_install role https://review.opendev.org/717717 | 19:03 |
| *** DanyC has joined #openstack-ansible | 19:14 | |
| jamesdenton_wfh | noonedeadpunk Re: https://bugs.launchpad.net/openstack-ansible/+bug/1859159... in my deployment, the services appear to have started OK as-is | 19:17 |
| openstack | Launchpad bug 1859159 in openstack-ansible "os_swift train incorect default systemctl configuration" [Undecided,In progress] - Assigned to Dmitriy Rabotyagov (noonedeadpunk) | 19:17 |
| jamesdenton_wfh | you referenced some CI errors? | 19:17 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/b0c73151991042c0bc51f6adb7b63baa/log/logs/host/swift-account-replicator.service.journal-12-59-05.log.txt | 19:19 |
| noonedeadpunk | it's not errors but it's weird | 19:19 |
| *** DanyC has quit IRC | 19:19 | |
| noonedeadpunk | in terms that it spawns process, replicator dies service terminates and re-spawns | 19:19 |
| noonedeadpunk | huh, this one actualy is ok | 19:20 |
| jamesdenton_wfh | not a respawn, just a 30-sec interval i guess? | 19:20 |
| jamesdenton_wfh | http://paste.openstack.org/show/791825/ | 19:20 |
| jamesdenton_wfh | thats my env | 19:20 |
| jamesdenton_wfh | granted, i don't know enough about Swift to know what is normal; | 19:21 |
| noonedeadpunk | jamesdenton_wfh: and your env launches swift-account-replicator ? | 19:23 |
| jamesdenton_wfh | yes | 19:23 |
| jamesdenton_wfh | http://paste.openstack.org/show/791826/ | 19:23 |
| noonedeadpunk | looks normal actually | 19:24 |
| noonedeadpunk | btw, another issue with this 1% | 19:25 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/b0c73151991042c0bc51f6adb7b63baa/log/logs/host/swift-container-replicator.service.journal-12-59-05.log.txt#286 | 19:25 |
| jamesdenton_wfh | lol | 19:26 |
| noonedeadpunk | yeah | 19:26 |
| noonedeadpunk | btw, I see actiually the same result for https://review.opendev.org/#/c/718524/2/defaults/main.yml | 19:27 |
| noonedeadpunk | https://b98e4b6eff505e4c0001-0ff94bb25fd1b2b22d207854a15b8765.ssl.cf2.rackcdn.com/718524/2/check/openstack-ansible-deploy-aio_metal-debian-buster/963e4e4/logs/host/swift-account-replicator.service.journal-18-59-56.log.txt | 19:27 |
| jamesdenton_wfh | weird. i checked the bins and they were indeed different | 19:29 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Fix replicator services https://review.opendev.org/718524 | 19:31 |
| noonedeadpunk | yeah... maybe both options just work... | 19:31 |
| jamesdenton_wfh | maybe so. i wish there were a bit more details in the bug report as to what the error was | 19:34 |
| noonedeadpunk | oh, wait | 19:35 |
| noonedeadpunk | you have swift-container-replicator.service | 19:35 |
| noonedeadpunk | while I guess we were discussing swift-container-replicator-server.serice? | 19:35 |
| noonedeadpunk | actualy I provided logs also for swift-container-replicator | 19:36 |
| noonedeadpunk | swift has soooo many services.... | 19:36 |
| noonedeadpunk | I think CI even don't have this one deployed | 19:37 |
| jamesdenton_wfh | http://paste.openstack.org/show/791827/ | 19:38 |
| jamesdenton_wfh | i don't have a swift-container-replicator-server.service | 19:38 |
| noonedeadpunk | uh | 19:40 |
| noonedeadpunk | so we have this one https://opendev.org/openstack/openstack-ansible-os_swift/src/branch/master/defaults/main.yml#L364-L370 | 19:41 |
| noonedeadpunk | and this one https://opendev.org/openstack/openstack-ansible-os_swift/src/branch/master/defaults/main.yml#L410-L415 | 19:41 |
| noonedeadpunk | looks like a mess | 19:42 |
| noonedeadpunk | And I think you don't have them because of disabled swift_dedicated_replication | 19:43 |
| *** kleini has quit IRC | 19:44 | |
| jamesdenton_wfh | ahh, i see. | 19:44 |
| jamesdenton_wfh | :facepalm: | 19:44 |
| jamesdenton_wfh | well, if it's broken, it's been broken a long time | 19:45 |
| jamesdenton_wfh | https://opendev.org/openstack/openstack-ansible-os_swift/commit/87fb727472b866aac31e7cbfdbc90d41f152cb03 | 19:46 |
| noonedeadpunk | yeah.... | 19:47 |
| noonedeadpunk | jrosser: do you recall smth from this? ^ | 19:49 |
| noonedeadpunk | I feel I really do need sandbox to understand what are we doing and why... | 19:52 |
| *** DanyC has joined #openstack-ansible | 19:53 | |
| noonedeadpunk | as in case of enabled swift_dedicated_replication it seems that we have 2 exactly same services with different naming | 19:54 |
| noonedeadpunk | jamesdenton_wfh: I guess that probably we should just use the same ternary in case of swift_dedicated_replication for swift-object-server | 19:56 |
| noonedeadpunk | that would make sense to me than.... | 19:56 |
| *** cshen has joined #openstack-ansible | 20:00 | |
| jamesdenton_wfh | seems redundant either way | 20:01 |
| jamesdenton_wfh | not what i mean, sorry | 20:01 |
| noonedeadpunk | jsut a sec | 20:02 |
| *** cshen has quit IRC | 20:05 | |
| noonedeadpunk | actually I'd say we could probably even combine account-server-replicator.conf.j2 and account-server.conf.j2.... | 20:05 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Fix replicator server services https://review.opendev.org/718524 | 20:05 |
| noonedeadpunk | does it looks fair at glance? | 20:05 |
| jamesdenton_wfh | at a glance | 20:06 |
| openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_swift master: Fix replicator server services https://review.opendev.org/718524 | 20:07 |
| *** gshippey has quit IRC | 20:07 | |
| noonedeadpunk | yeah... | 20:08 |
| noonedeadpunk | maybe jrosser has some extra expertice with swift... | 20:08 |
| jamesdenton_wfh | hopefully! | 20:09 |
| jamesdenton_wfh | look at the 1% issue. i have a theory, waiting for playbooks to finish | 20:09 |
| noonedeadpunk | oh, nice | 20:10 |
| noonedeadpunk | jamesdenton_wfh: since you're here, can you probably vote on https://review.opendev.org/#/c/717721/ ? | 20:11 |
| noonedeadpunk | I think CI is failing because of this now... | 20:12 |
| jamesdenton_wfh | url and checksum look OK | 20:13 |
| jamesdenton_wfh | done | 20:13 |
| noonedeadpunk | thanks | 20:13 |
| jamesdenton_wfh | noonedeadpunk my theory is bunk | 20:24 |
| jamesdenton_wfh | anyway, the 1% issue doesn't seem to affect this build: https://zuul.opendev.org/t/openstack/build/bfd3249a28dc481294afbbfe89336376/log/logs/host/swift-account-server.service.journal-21-41-40.log.txt | 20:24 |
| jamesdenton_wfh | https://review.opendev.org/#/c/718208/ | 20:24 |
| timburke | what's this 1% issue? sounds like https://bugs.launchpad.net/swift/+bug/1844368 -- what version of swift is it? | 20:35 |
| openstack | Launchpad bug 1844368 in OpenStack Object Storage (swift) "fallocate_reserve cannot be specified as a percentage running under python3" [High,Fix released] | 20:35 |
| jamesdenton_wfh | that's the one | 20:39 |
| jamesdenton_wfh | i believe it was a recent os_swift master build | 20:40 |
| jamesdenton_wfh | i am not entirely sure if it's a case of fallocate_reserve = "1%" (doesn't work) or fallocate_reserve = 1% (works) | 20:41 |
| timburke | fwiw, i'd expect that second one -- idk that i've ever messed with trying to have quotes in swift config files... | 20:48 |
| jamesdenton_wfh | i'm not sure of the conditions of the failure. will have to wait for noonedeadpunk to confirm. | 20:56 |
| *** thuydang has quit IRC | 21:26 | |
| *** thuydang has joined #openstack-ansible | 21:49 | |
| *** cshen has joined #openstack-ansible | 22:01 | |
| *** cshen has quit IRC | 22:05 | |
| *** krah has quit IRC | 22:09 | |
| *** gyee has quit IRC | 22:49 | |
| mnaser | jrosser: ouch, that sucks | 22:56 |
| *** threestrands has joined #openstack-ansible | 23:10 | |
| *** tosky has quit IRC | 23:20 | |
| *** DanyC has quit IRC | 23:25 | |
| thuydang | Hi, I could install using OpenStack-ansible and have error creating instance. For troubleshooting an anyone tell me where all the service-log files are? I search /openstack/log/infra-xx on log-hosts but did not find them. Thanks | 23:25 |
| *** gary_perkins has quit IRC | 23:26 | |
| *** gary_perkins has joined #openstack-ansible | 23:32 | |
| thuydang | Ah, I can see the log using journalctl inside the lxc container. Is that the right way? | 23:55 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!